tweetfilter.rb

# Simple message filtering logic for a Ruby-based Twitter client.
#
# This demonstrates two simple string-based filtering approaches: 
# * blacklist words
# * blacklist phrases
#
# It uses a slightly optimized matching and shows a sub-optimal one as comment.
#
# This can ofcourse also be used for highlights, dynamic grouping and whatnot.
messages = [
  'First tweet',
  'Foursquare annoys me',
  'I am awesome #foursquare',
  'I twote about tweeter'
]

filter_words = ['#foursquare', 'tweet']
filter_phrases = [
  'twote about'
]


# Runs in O(msg_words * filter_words):
#   leftover_words = messages.reject {|msg| msg.split(" ") & filter_words).size > 0}

# Terminates early as soon as a match is found.
# Runs in O(msg_words * filter_words) only in worst case (no match or last words match)
# Average case is O(0.5 * msg_words * filter_words)
leftover_words = messages.reject do |msg|
  words = msg.split(" ")
  words.find {|msg_word| filter_words.find {|filter_word| msg_word == filter_word}}
end

leftover_phrases = messages.reject do |msg|
  # Uses Array#find to return a true value as soon as a match is found
  filter_phrases.find {|phrase| msg.include?(phrase)}
end
leftover_both = leftover_words & leftover_phrases

p leftover_words # => ["Foursquare annoys me", "I twote about tweeter"]
p leftover_phrases # => ["First tweet", "Foursquare annoys me", "I am awesome #foursquare"]
p leftover_both # => ["Foursquare annoys me"]